A few years ago, on a Delta flight, I noticed that the airline was boarding people in the middle first. I asked the flight attendant about it, and he said that studies had shown that it was faster than back to front, which surprised me, because the latter had always been conventional wisdom and industry practice. Now, American claims that, based on simulations, random boarding is better yet. I’d be interested to see a plausible explanation for this, if true.
23 thoughts on “Non-Intuitive”
Comments are closed.
I have one theory for the middle before back. Those people who have back of the plane seats that put their carry on baggage in overheads near the middle of the plane. Part of this is caused by airlines/manufacturers having overhead bins with inflatable rafts, first aid kits, blankets, and pillows in the last few bins. Thus giving frequent travellers of the back reason to use middle bins. People sitting in the middle are less likely to worry about not having a bin.
Another benefit is once the middle is filled, you have a buffer for loading people into the back. So what may seem random is actually smoothed by the buffer. Then again, the randomness of open boarding would allow the concourse to serve the purpose of boarding. After all, it takes a pretty dense fellow not to notice the herd trying to pack themselves down the jetway.
What I wonder is how much fees for checked baggage motivates people to take more as carry on, thus negatively affecting boarding times.
I always thought it would be faster if they loaded window seats first and then aisle seats. Less waiting while people had to get up and let others in.
I’ll give it a whirl.
When you board in a block, say rows 30-35, you have 6×5 = thirty people trying to cram their gear into the ~30 available overhead spots. First, they’re all blocking one another attempts. “I’ve reached my seat, but I can’t shove anything into the overhead until he sits.” Second, when they do get close to full, it becomes easier to ‘overflow’ the allotted area than to seek and find the remaining spot or wedge it into a spot ‘almost’ big enough. So – the next block has an even worse time than the first. And the last block becomes ‘search the plane for an empty overhead’. The last couple slots are generally -well- hidden, because the overhead bins were basically filled like pouring sand in a cup with very few gaps. But it isn’t -tightly- packed, because few individuals ‘fought’ to get their gear into their ‘optimum spots’
Random boarding is obviously adding the congestion on the aisle. But the ease of just shoving your bag into the overhead a row a head or two is hampered. With the plane half full, most people managed to put their bags directly over their heads. So the last people filling in are all fighting their bags into cramped positions, shifting inconveniently placed bags, etc.
IOW: It’s because we’ve shifted from “getting in” being the time consuming piece to “packing the plane” being the time consuming piece. Because $25 baggage fees mean every single person has a near-maximal carryon bulging at the seams.
PS we should charge the airlines union baggage packer’s rates for the work. /jk
I always thought it would be faster if they loaded window seats first and then aisle seats.
As the article mentions, some airlines have been doing that (which I like, being a window seater). But American is claiming that random is better still.
I wonder if anyone’s ever tried odds-n-evens…
Here’s a thought: set up the jetways to allow boarding through the rear doors as well as the front.
One of my biggest annoyances with flying is when people with rear seats put their carry-on bags up front because they’re too lazy to carry them to the back. I’ve had that happen several times but fortunately, some flight attendants are trying to stop it.
I love airplanes more than most people (which can be read two ways and they’re both accurate). Still, the airlines have sucked the joy out of flying. In late May through early June, I had to make two trips from Colorado Springs to Huntsville, AL due to a family emergency. I drove both times (about 2600 miles round-trip). The last minute airfare would’ve been $1000 each time and it just wasn’t worth the hassle or the expense.
I swear I saw a study on this a few years ago, including that random boarding was indeed faster, possibly with computer animations and simulations of networks of people to demonstrate the theory.
Either that, or it was a story about how un-assigned seating led to quicker boarding, because people just naturally spaced themselves out and didn’t all try to get into the same overhead bin at the same time. Seems that unassigned seating would be just about as close to “random boarding” as anything else.
Here’s a thought: set up the jetways to allow boarding through the rear doors as well as the front.
Damn you and your practicality. It’s some nitwit brainiac like you that gave us jetways in the first place; replacing airstairs and mobile stairways.
Either that, or it was a story about how un-assigned seating led to quicker boarding, because people just naturally spaced themselves out and didn’t all try to get into the same overhead bin at the same time.
Southwest has used unassigned seating for years. On the few occassions when I’ve flown with them, it seemed to work well. Just be sure to log online as early as possible so you get a low boarding number. They don’t charge for checked luggage and that may be a big part of it. People don’t feel the need to bring everything they need for a trip on board as carry-on luggage.
At my local ‘port, passengers walk out onto the tarmac to and from the plane, so this is exactly what happens.
It makes perfect sense. The long pole isn’t the aisle space, it’s the final few feet to get luggage put away and seated. If there’s less congestion around individual seat areas then it’s much faster to stow your luggage and sit down.
My first time getting on an airplane, I was profoundly disappointed that we went via a jetway instead of up a stairway on the tarmac like I always saw in movies or on TV.
MY Theory? They are using the blood flow model of traffic transfer, assuming that congestion actually increases transfer. It’s an old idea that has been used to rationalize a lack of road expansion. I believe it goes back to ’94. I read it in discover back then.
The other? Block allowances don’t matter, because most people figure “Eff it” and they wait until the line shrinks before boarding anyways, simply cuz they know where they are gonna sit, and can just ask a steward/attendant to find a location for their carryon anyways. So block seating is, in effect, random seating anyways because the passengers self select to be a part of general seating.
Ok… wait for it… hammocks! Planes with no seating or luggage racks at all. No seats to restrict movement onto the plane. One end of the hammock is permanently attached to each side of the plane. When the other end is hooked a center isle remains. Ok, I’m weird. I just like sleeping on planes.
Luggage fits in bags and hooked to the sidewall. Top hammocks board first.
What does it matter if they find the most efficient/fastest way to pack people on the plane, if they then have to sit waiting on the tarmac for several hours?
Ed, what a party pooper. For my next non intuitive I was gonna go with water slides…
Here’s an argument that random is a little worse in terms of average boarding time.
Suppose the plane has one seat each row with seats labeled 1 to N. When we ask N customers to board we assume they board randomly. That is they sort themselves into a random permutation. There’s a delay when the seat assignment in the permutation is lower than the next seat assignment in the permutation (this is a definition, I know there are other ways we could decide a delay occurs but this is easy). Let D(N) be the average number of delays that occur. It’s not too hard to show that D(N) = (N-1)/2. Now a random boarding pattern of N*M people will be better than boarding in M blocks of N if D(N*M) < M*D(N). But D(N*M) = (N*M-1)/2 and M*D(N) = M*(N-1)/2 so it appears that block boarding is somewhat better.
On the other hand, I suspect that the standard deviation of the delay actually does go down when you move from the block system to the single random permutation. Which means that the schedule incurs less risk.
If this is true, then while the random system is not more convenient for the customer on average, it may be better for the airline since it makes the scheduling less volatile.
Robin Goodfellow’s explanation seems perfectly logical, and Larry J already referenced Southwest. I’ll just add that these days I mostly fly Southwest, and it is very much a random system. Although it follows a weird preference numbering system as far as who gets to board when, I’ve noticed that even when I’m part of the “A” group, many of the A’s prefer to sit up front, but the rest of us spread ourselves out all over the place.
It’s a confusion of terms. It’s not random in the sense that people sit anywhere they like. They’re still assigned to a particular seat. It’s “random” in the sense that the people assigned to board are randomly selected from the people destined to board, and who all have assigned seats already.
As I understand it from the article, the reason is straightforward: the biggest cause of delay is people getting in each other’s way because they arrive at closely spaced seats at about the same time and are all trying to do the same things — stuff luggage in overhead bins, find another bin, sit down, stuff a cat under the seat, whatever.
So all you really want is some method that keeps people simultaneously reaching their seats as far apart as possible. There are certainly algorithms that will do that, but it would be a pretty complex calculation that would have to be optimized for each different plane seating arrangement, exact distribution of passengers assigned seats, taking into account any passengers who have paid for advanced boarding, et cetera.
And you can probably get pretty close by just randomly assigning people a boarding order, because the probability that any two people reaching their seat simultaneously would be right next to each other is relatively small, of order (n/ns)², where n = number of people and ns = number of seats.
Here’s a way to think about that: take a map on paper of the airplane, and color all the seats to which passengers have been assigned.
Now start randomly drawing X’s on the colored spots. How often are you going to draw an X right next to the previous seat on which you drew an X? Not often.
There are certainly ways to draw the X’s so that each choice is as far away from the previous choice as possible, but that’s a complex calculation.
Doesn’t have to be totally random to get the same benefits. Just give a different color boarding pass for each third of the plane. Let anyone board that doesn’t have the same color as the person in line in front of them.